Watching the Waist of IETF 51 the Steve London Deering Protocol August deering@ 2001 Hourglass cisco.com
1 email WWW phone...
SMTP HTTP RTP...
TCP UDP…
IP
ethernet PPP…
CSMA async sonet...
copper fiber radio...
2 Why the Hourglass Architecture?
6 Why an internet layer? • make a bigger network • global addressing email WWW phone... SMTP HTTP RTP...
• virtualize network to isolate end-to-end TCP UDP…
protocols from network details/changes IP
6 Why a single internet protocol? ethernet PPP… CSMA async sonet... • maximize interoperability copper fiber radio... • minimize number of service interfaces 6 Why a narrow internet protocol? • assumes least common network functionality to maximize number of usable networks 3 Why Am I Talking About Watching the Waist? 6 Plenary talk is an opportunity for navel gazing 6 It happens on reaching middle age (me & IP) 6 The IP layer is the only layer small enough for me to get my arms around 6 I am worried about how the architecture is being damaged: the waste of the hourglass 6 The hourglass theme offers many bad puns
4 Putting on email WWW phone... SMTP HTTP RTP...
Weight TCP UDP…
IP + mcast
+ QoS +...
ethernet PPP… • requires more CSMA async sonet... functionality copper fiber radio... from underlying networks
5 Mid-Life Crisis email WWW phone... SMTP HTTP RTP...
TCP UDP… • doubles number of service IP4 IP6 interfaces ethernet PPP… • requires changes CSMA async sonet... above & below
copper fiber radio... • creates interoper- ability problems
6 Oops! email WWW phone... • NATs & ALGs An used to glue the SMTP HTTP RTP... broken pieces Accident TCP UDP… • lots of kinds of IP new glue being invented—ruins predictability ethernet PPP… • some apps CSMA async sonet... remain broken, copper fiber radio... since repairs are incomplete
7 Threat- • danger : creeping ened by email WWW phone... dependencies on SMTP HTTP RTP... specific link- layers inhibit Youths TCP UDP… flexibility and evolution IP ATM • will never fully ethernet supplant IP, so end up with CSMA async sonet... complicated copper fiber radio... hybrid & more address plans
8 But Still Supple email WWW phone... SMTP HTTP RTP...
TCP UDP… IP • IP-over-IP tunneling has become more and ethernet PPP… more common CSMA async sonet... • this is not so bad: copper fiber radio... retains benefits of hourglass model
9 More Fattening Temptations
6 TCP “helpers” 6 reliable multicast assists 6 packet-intercepting caches
6 “content-based routing” email WWW phone... SMTP HTTP RTP... 6 active networking TCP UDP… IP + ? + ? + ? +...
ethernet PPP…
CSMA async sonet...
copper fiber radio...
10 Lost Features of the Internet
6 transparency 6 robustness through “fate sharing” 6 dynamic routing 6 unique addresses 6 stable addresses 6 connectionless service 6 always-on service 6 peer-to-peer communication model 6 application independence
11 Below-the-Waist Bulge
6 mostly reinventing, badly, what IP already does (or could do, or should do): • VLANs • LANE (LAN emulation / “interworking”) • layer 2 tunneling protocols • MPLS, PPPoE,… (“layer 2.5”) 6 lower layers mostly seem to just make IP’s job harder • cells, circuits, QoS, multicast, large clouds, opaque clouds
12 What to Do?
6 First, acknowledge that this is the normal entropy / decay that besets all large, engineered systems over time 6 So, shall we just let nature take its course? 6 Or, shall we make the effort to get back into shape?
13 A email WWW phone... Fitness • perhaps we can SMTP HTTP RTP... trim down from Goal TCP UDP… an hourglass to a wineglass
IP6 • promising signs: IP-over-SONET, IP-over-WDM • IPv6 to restore simplicity and copper fiber radio... functionality
14 Only Time Will Tell…
15